Need Help with Text to Speech

[dothedog]

Just for background I have a linuxMCE 710b4? system up and running for a while now. I have 6 media directors that all work for playing music in each room. The core is a hybrid in the home theater. What I want to do is send a text to speech message to my kids rooms to wake them up in the morning . Here is what I did:

I followed the instructions here

I went to the Core, and made sure that Text To Speech was checked. I went to Wizard-->Scenarios-->Lighting Scenarios and added a scenario to the Living Room (that is where I am when I want to wake the kids) Edit the scenario and change the name to "Kids Wakeup Call", hit update, change edit scenario to advanced wizard, from drop-down select "Send Audio to device". In #103 List PK Device (string), tried the device numbers for the mplayer on the orbiter for the kids rooms on-screen media director, and in #9 Text (string) typed in "wake up now" without the quotes, then hit save.

When I hit the "Test" button, nothing happens. Looking in the log, I get this:

Code: [Select]

Wed Jan  7 06:32:51 MST 2009 Restart
========== NEW LOG SECTION ==========
1       01/07/09 06:32:59       18 (spawning-device)    Starting... 1
1       01/07/09 06:32:59       18 (spawning-device)    Found /usr/pluto/bin/Text_To_Speech
05      01/07/09 6:33:01.091            Connect() failed, Error Code 111 (Connection refused)) <0xb799f6c0>
05      01/07/09 6:33:02.091            Connect() failed, Error Code 111 (Connection refused)) <0xb799f6c0>
05      01/07/09 6:33:03.091            Connect() failed, Error Code 111 (Connection refused)) <0xb799f6c0>
05      01/07/09 6:33:04.091            Connect() failed, Error Code 111 (Connection refused)) <0xb799f6c0>
05      01/07/09 7:35:25.871
One other thing that looks a little weird is, in the Commands box in the scenario, it has this: Text To Speech [Home Theater], Send Audio to Device. Is that right? I thought is was coming from the core, but because my core is a hybrid in the Home Theater, is it coming from the MD on the core?

I guess the other question I have is which device should I send the Text to Speech to? I have tried mplayer on the on-screen Orbiter for the rooms, xine, and even the MD device number. None seem to work...

Am I missing something?

Any and all help is appreciated!!!

Dothedog

[wombiroller]

I had not used this before but just tested it out. The "Test" Button doesn't work for me either (not sure why), however, try regenerating the an orbiter and the running the scenario from the generated button (e.g. Under Lighting which is were I put mine).

Works for me - but only after regenerating the orbiter. Same goes for any changes to the text in #9 Text (string). These work - but only after you regen the orbiter...

Cheers,
WR.

[jondecker76]

its been a while, but I seem to remember that the test button would work if the text is enclosed in quotes

[wombiroller]

You're correct Jon!

Single quotes that is - double quotes work for the testing button also - but then you can't save the values....

Cheers,
WR.

[jondecker76]

Ok, I thought I remembered that...
thats a simple web admin bug - i'll look into fixing it this weekend

[dothedog]

Thank you for the responses. It looks like it only works if you put the device# for mplayer in there. The single quote also helped. I was wondering why it wouldn't save with double quotes. The next problem is that in the #9 Text (string) field I typed 'wake up now' for some reason it only says "wake". Is there a trick to this too?

[jondecker76]

I think the single quotes work just for the test button. Take them back out when you are happy with the sound and you are ready to save the command (I'll patch this up soon in the web admin)

I'm quite sure that you can send the on-screen orbiter device number for text to speech - give it a try.

[dothedog]

jondecker76,
Thanks for the reply. After playing around with it a little, you are correct, you can put the device number for the on-screen orbiter in there. The text works either with single quotes or no quotes from the scenario button. I am having some problems though. I am sending this to three rooms. In two of the rooms it only says "wake" in the third room it says the whole thing "wake up now". Any ideas on why that is?

Also, it may be good to clarify in the wiki to use the on-board orbiter device number. It wasn't clear to me when I read it.

Thanks again for your help!

DoTheDog

[bulek]

jondecker76,
Thanks for the reply. After playing around with it a little, you are correct, you can put the device number for the on-screen orbiter in there. The text works either with single quotes or no quotes from the scenario button. I am having some problems though. I am sending this to three rooms. In two of the rooms it only says "wake" in the third room it says the whole thing "wake up now". Any ideas on why that is?

Also, it may be good to clarify in the wiki to use the on-board orbiter device number. It wasn't clear to me when I read it.

Thanks again for your help!

DoTheDog

Hi,

I'm also spotting improper behaviour with speech announcements - specially with missing parts of words spoken... It seems that text to speech is currently not handled properly (you can try sending more texts at once and it will choke). It also doesn't make right countdown on arming alarm....
It all shows like it doesn't make any queue out of announcements that are coming in too fast...

I think we don't have much developers working on that - so I'm event thinking about writing Text_To_Speech device in Ruby, so it can be easily tested and extended....

I'm also trying to replace Flite with Festival that has much better voices, but it seems that Flite is hardcoded....

Regards,

Bulek.

[tschak909]

ew, do _NOT_ write the TTS in ruby!

-Thom

[bulek]

ew, do _NOT_ write the TTS in ruby!

-Thom

Why not? Plugin basically runs external flite application that makes wave file and .... I guess maybe being plugin is a drawback, but if I understand right, there are no time critical part in it...

Anyway it was just a thought.... What is needed is that we make our TTS plugin a bit smarter. It could save announcements that are frequently used to files. Also it would be fine to allow different TTS engines.... Festival is surely viable alternative, since we have quite powerful core...

What I'm particularly missing is that even Security countdown doesn't perform well and also if many speech announcements are scheduled that they appear incomplete with missing parts.... Does anyone know where this happens - in player or in synthesized waves ?

Regards,

Bulek.